Cash holdings and the financial crisis: Evidence for MNCs in developed countries

Cash holdings and the financial crisis:

Evidence for MNCs in developed countries

Name: Julia Goldhausen Student number: s3494209

Study program: MSc IFM Faculty of Economics and Business, University of Groningen Supervisor: Dr. P.P.M. Smid

Co-Assessor: Dr. H. Gonenc

Abstract: This study investigates the impact of the pre-crisis (2003-2007), crisis (2008-2011) and post-crisis period (2012-2016) on the cash holding behavior of a large sample of listed and unlisted multinational companies from 31 developed countries, thereby also controlling for the impact of access to finance as well as the legal origin, using panel data. The results indicate that companies increase their cash holdings during the crisis and to a further extend in the post-crisis time. The precautionary motive can be one explanation for this behavior as the availability of financing sources is more restricted from 2008 onwards. Multinationals in common law legal systems hold, on average, less cash than firms in code law legal systems.

Keywords: cash holding, financial crisis, access to finance, MNCs, developed countries, effect of the legal system

1. Introduction

The reasons for the recent global financial crisis are diverse. These include, for instance, subprime mortgage defaults, credit booms as well as the housing boom (Acharya et al., 2009). Although the crisis commenced in the United States in 2007, it gradually enlarged and by 2008 affected countries worldwide (Acharya & Richardson, 2009). Economic downturns like the recent financial crisis also have an impact on the consumption of society and thus, ultimately affect companies as well.

In perfect markets without transaction costs the amount of cash a company has disposable is irrelevant. However, these markets do not exist in reality. Rather, markets are imperfect as, for example, transaction costs occur. In addition, problems of foregoing investment opportunities due to the non-availability of funding may arise, in case a firm does not keep an adequate amount of cash obtainable (Keynes, 1936; Opler et al., 1999). The latter can be of more concern in times of crises when organizations become more financially constrained due to lower expenditures of society leading to lower cash inflows. Furthermore, corporations may also have restricted access to bank financing. Therefore, the recent financial crisis gives a unique opportunity to study the cash holding behavior of firms all being impacted by an economic shock. Moreover, the aim of this study is to provide new evidence for this period and analyze how multinational companies (hereafter referred to as MNCs) in different developed countries, as well as, in distinct stages of the business cycle should cope with external shocks affecting their daily business. Although MNCs account for merely 2% of employment worldwide, these corporations are at the head of supply chains controlling about half the global trade (The Economist, 2018). Consequently, MNCs are very important for the global economy. The focus is on firms in developed countries as these differ from corporations in emerging markets. For instance, Pinkowitz et al. (2015) find evidence that organizations in the developed world have lower cash holdings compared to companies in developing countries. The key motivation for this research is to provide new evidence for the cash holding behavior during the crisis for a large international sample of listed, as well as, unlisted firms and thereby also controlling for the impact of access to financing and the legal origin of regions, namely the common as well as the code law legal system.

countries given their activity in multiple areas around the world (Jang, 2017). Therefore, these corporations should, in general, have less need for precautionary cash savings (Fernandes & Gonenc, 2016). However, uncertainty increases during times of crises and it is to be investigated if the cash holding behavior of MNCs changes among the periods under investigation, namely the pre-crisis (2003-2007), crisis (2008-2011) and post-crisis time (2012-2016), given the perceived uncertainty. Therefore, the precautionary motive as described by Opler et al. (1999) may become of concern for firms being active internationally, as countries worldwide are affected by the recent financial crisis. In turn MNCs may increase their cash holdings to pursue investment opportunities. This study may, consequently, help to establish corporate policies in order to cope with external environmental changing forces which impact the capital structure of these companies.

Recent studies on the cash holding behavior of corporations provide evidence that the amount of cash companies keep at hand is increasing (Amess et al., 2015) and so is the amount of literature on cash holding. Bates et al. (2009) confirm an increase in cash holdings for US firms between 1980 and 2006 and reason that one explanation for this behavior is the precautionary motive. A recent study by Kahle and Stulz (2013), also investigating a US based sample of corporations, concludes that companies which were reliant on bank financing before the offset of the financial crisis increase their cash holdings during the crisis. In contrast, non-levered firms decrease their cash holdings in the years after the crisis outbreak. However, it is still to be investigated whether this relationship also holds across countries. The recent financial crisis influenced business operations substantially and is hence an important and interesting factor to study. Additionally, the cash holding behavior, as well as, being financially constrained due to restricted access to financing are crucial management concerns. For financially constrained firms remaining competitive and ensuring a smooth flow of everyday business is challenging. Most studies on the cash holding behavior of organizations do not account specifically for the time of crisis in comparing changes of firms’ cash holdings between periods of time (e.g. Opler et al., 1999; Bates et al., 2009). In addition, the focus of previous researches is mostly on publicly listed firms (e.g. Opler et al., 1999; Almeida et al., 2004) as well as on single countries (e.g. Bigelli & Sánchez-Vidal, 2012; Kling et al., 2014). Moreover, the majority investigates the cash holding behavior of firms incorporated in the United States (e.g. Opler et al., 1999; Bates et al., 2009; Kahle & Stulz, 2013).

changes between the pre-crisis, crisis and post-crisis period. Due to the fact that enterprises become more financially constrained in terms of obtaining financing during the crisis.

Accordingly, the main research question to be answered in this study is the following: Does the cash holding behavior of MNCs incorporated in developed countries differ during the recent financial crisis from the cash holding behavior during the pre- and post-crisis years? Furthermore, this study aims to answer the following two sub-questions:

Does having easy access to financing impact the cash holding behavior differently during the period of pre- and post-crisis compared to the years of crisis?

Does the legal origin of the country the firm is incorporated in have a different influence on the cash holding behavior during the period of pre- and post-crisis compared to the years of crisis?

This study uses panel regressions and includes a large, international sample of MNCs from 31 developed countries with 81,935 firm year-observations for the period of 2003 until 2016. To provide evidence for the second sub-question, i.e. the impact of the legal origin, the initial sample is modified by reducing it to companies being incorporated in Germany and Japan, as well as firms being founded in Great Britain and the United States to account for countries with a code and common law legal system, respectively. Given the fact that these countries are considered to be the major countries fitting each respective category.

pre-crisis compared to the years of crisis, whereas it is more pronounced in the post-crisis years compared to the time of crisis. One explanation for this behavior may be agency theory. The paper proceeds as follows: first, in section 2 the literature is reviewed, and the hypotheses are developed, followed by a section about the methodology and the data used in this research. Then the empirical findings are outlined in section 4 and finally a conclusion is provided. 2. Literature review

2.1. Theoretical background

Holding cash has the advantage that investment opportunities can be pursued directly, without the need of taking on debt or issuing equity. However, keeping cash at hand is still costly. Reasons for this include liquidity premiums (Opler et al., 1999) as well as tax disadvantages (Faulkender & Wang, 2006). Relating to this, Faulkender and Wang (2006) find that there may be a maximum amount of cash holdings that create value for firms, and corporations holding more cash than this optimum destroy value. Hence, cash can, in this case, be considered as financial slack since it is not used for value creating purposes in a company. Thus, there is a tradeoff in cash holding and therefore, the cash holding behavior of businesses may be different within the period of crisis compared to the pre- as well as the post-crisis time.

According to previous literature, the cash holding behavior of firms is determined by several factors. First, the transaction cost motive being concerned about raising external capital which comes at a cost is considered. Following Keynes (1936) and Opler et al. (1999), corporations having assets that can be easily converted into cash or creating cash by skipping dividend payments to shareholders, have lower transaction costs than firms that cannot easily substitute their need of capital. Although accessing the capital market is preferred as it is less costly than converting assets into cash, fixed costs are encountered when accessing the capital market as well. Uncertain cash flows are another concern under the transaction motive. The higher the uncertainty of cash flows, the more cash is hold back to compensate for times of lower cash inflows. Therefore, this motive can be of importance during the time of crisis when cash flows become more uncertain given the economic downturn and higher fluctuations in consumer spending, as well as, restricted access to external financing. In more general terms, enterprises become financially constrained as they do not obtain the financing they desire to pursue investment opportunities.

available (Opler et al., 1999). Given the negative economic shocks of a financial crisis, it is thus expected that firms increase their cash holdings during the time of crisis, because of the precautionary motive, to compensate for the constrained access to financing.

Under the agency motive, i.e. the potential conflict of interest between the managers and the owners of a corporation, the cash holding behavior of managers may be non-optimal. Managers may hold higher amounts of cash than ideal due to the fact that they can pursue investment opportunities, in case cash is readily available, giving them more flexibility (Jensen, 1986). However, these investments may be sub-optimal since managers can engage in empire building, namely growing the enterprise beyond optimal (Hope & Thomas, 2008). From a shareholder perspective large amounts of cash holdings may thus not be desirable as a more than optimal level of cash is value destroying for the company as demonstrated by Faulkender and Wang (2006). Therefore, it can be argued that shareholders prefer large amounts of cash being paid out via share repurchases, as interest income of cash is taxed twice (i.e. once at the corporate level and once at the personal level). Shareholders benefit from this since cash holding costs increment with the marginal tax rate of a corporation. In case of a share repurchase the marginal tax rate decreases by the corporate tax rate (Opler et al., 1999). Furthermore, with significant amounts of cash managers cannot only pursue investment opportunities without the need of raising external capital, they also do not need to find arguments for their investment opportunities to convince potential investors or banks (Jensen, 1986; Opler et al., 1999). This in turn may lead to investments in negative net present value projects that are value destroying for the company due to the fact that the accumulated cash needs to be spend quickly (Opler et al., 1999). Additionally, there are opportunity costs related to high amounts of cash holdings, and it can be expensive for corporations to hold cash due to the fact that liquidity premium returns are low (Acharya et al., 2013).

is more pronounced for firms experiencing high costs in accumulating financing, mainly those being financially constrained which is a greater issue during times of economic downturns. For these companies, it may be cheaper to hold cash instead of trying to get a loan or liquidating assts (Acharya et al., 2013).

Furthermore, from the pecking order theory perspective, corporations prefer to use internal financing over external financing as internally generated resources are cheaper due to the fact that no information asymmetries exist. Therefore, the involved risk is also lower (Myers, 1984). If external financing is needed, debt is preferred over equity as the perceived risk of equity financing is higher than the one for short-term and long-term debt financing. This can be explained by the fact that asymmetric information can lead to new stock issues being undervalued. Thus, the premium asked for equity financing is higher compared to the one for debt. Hence, equity financing is less preferred than debt financing from the pecking order perspective (Myers & Majluf, 1984). Taking the preference for internal financing into account and the fact that external financing can be expensive and unavailable during times of crises, higher cash inflows should lead to higher cash holdings.

Firms established in code and common law legal systems differ in more terms than only the financing source used. Countries with a common law system are considered to have higher outside investor protection mechanisms compared to countries with a code law legal system (Doupnik & Perera, 2015). Furthermore, La Porta et al. (2000) emphasize that the legal system can serve as a proxy for corporate governance standards within different countries, given the fact that legal rules are the foundation for the preferred financing source. More specifically, in countries with a common law system, legal rules are made by judges based on precedents and this led to higher investor protection of outsiders, no matter of their size, and thus a tendency for equity financing. On the other hand, countries based on code laws adopted a civil system in which laws are made by the government and need to be followed. This resulted in a more investor protected focus of concentrated ownership, and therefore bank financing serves as the main source of funds (La Porta et al., 2000).

2.2. Previous empirical findings

2.2.1. Cash holdings and the crisis

Due to economic shocks cash inflows of corporations during the time of crisis may be lower, leading to companies being more financially constrained. Relating to this, the transaction cost as well as the precautionary motive give rise to the expectation that corporations hold more cash due to higher uncertainties which can be compensated by an increase in cash holdings (Keynes, 1936; Opler et al., 1999).

Furthermore, Chang et al. (2017) conclude for a sample of US companies that enterprises increase their cash holdings in the period of crisis compared to pre-crisis years. However, the effect is stronger for financially constrained firms than for unconstrained corporations. These findings are supported by Almeida et al. (2004) who find evidence for a sample of manufacturing enterprises that financially constrained organizations hold more cash than unconstrained ones. Empirical evidence of Kahle and Stulz (2013), using quarterly data of US firms, shows a decrease in cash holdings after the beginning of the financial crisis compared to the pre-crisis years. However, in the last year of the financial crisis (according to their definition quarter two, 2009 until quarter one, 2010), corporations are able to increase their cash holdings to a level similar to the one of the pre-crisis period. Moreover, Song and Lee (2012) demonstrate that the precautionary motive is of importance during crisis times. In their sample of East Asian firms, enterprises increase their cash holdings after the offset of the Asian financial crisis in 1997-1998. Due to the experience of negative economic shocks, these corporations change their cash holding policies in the long-run in favor of higher amounts of cash holdings.

Based on these former findings, it is expected that firms increase their cash holdings during the time of crisis compared to the pre-crisis period to avoid the forgoing of potential investment opportunities. Although only the study of Song and Lee (2012) investigates the differences in cash holdings during the post-crisis years, following their evidence, it can be expected that also MNCs in developed countries hold higher amounts of cash in the post-crisis period compared to the years of crisis.

Hypothesis 1a: MNCs incorporated in developed countries have lower cash holdings during the pre-crisis years compared to the years of crisis.

2.2.2. Impact of cash flow on cash holdings

D’Mello et al. (2008) emphasize for their sample of US enterprises during the period of 1985-2000 that organizations having easy access to internal financing in terms of higher cash flows, tend to hold lower amounts of cash. On the other hand, Ozkan and Ozkan (2004) find evidence for a sample of publicly listed firms in the UK from 1984 to 1999 that cash flow is positively related to cash holdings. This is confirmed by Song and Lee (2012) for a sample of East Asian corporations, as well as Opler et al. (1999), who also find a positive relationship among cash flow and cash holdings for publicly listed organizations in the US. Additionally, Chang et al. (2017) report a positive influence of cash flow on cash holdings during the timeframe of 2002-2010, as well. Furthermore, following the pecking order theory, internal financing is cheaper compared to external financing, predicting a positive relationship, too.

Although seen desirable from the precautionary as well as the transaction cost motive, the question remains whether cash holdings can indeed be increased in case the organization’s cash inflows are lower in times of crisis. In addition, evidence from Denis and Sibilkov (2010) demonstrates that US public firms with lower cash inflows also have smaller amounts of cash holdings due to the fact that these corporations are unable to build cash reserves given their limited amounts of cash flows from 1985-2006. Opler et al. (1999) maintain for a sample of US public companies that organizations with unstable cash flows hold higher amounts of cash. Relating these findings to the impact of the recent financial crisis, it is reasoned that managers save higher amounts of cash inflows as cash, thereby increasing their cash holdings, to compensate for times of more restricted access to financing. Evidence from Han and Qiu (2007) underlines this prediction. In their sample of US publicly listed corporations, from 1997 until 2002, financially constrained firms, increase their cash holdings as cash flows become more uncertain. Relating to this, international evidence of Song and Lee (2012) confirms the former results. In their sample of East Asian corporations, enterprises with higher cash inflows increase their cash holdings during the Asian financial crisis. In turn, before and after the economic downturn cash flows are more predictable. Hence, it is reasoned that cash holdings stemming from cash inflows are lower in the two non-crisis periods. Therefore, the next hypotheses are as follows:

Hypothesis 2: For MNCs incorporated in developed countries, cash flow has a positive impact on cash holdings.

Hypothesis 2b: The relationship of cash flow and cash holdings is less pronounced during the post-crisis period compared to the period of crisis.

2.2.3. Impact of easy access to bank financing on cash holdings

Bigelli and Sánchez-Vidal (2012) conclude from a pre-crisis sample of private Italian companies that bank financing can serve as a substitute to high amounts of cash holdings. Therefore, firms having easy access to bank financing, can use newly issued debt instead of keeping large amounts of cash at hand to execute investment opportunities. This is confirmed by Opler et al. (1999) who reason for a US based sample in the period of 1971-1994 that firms with better access to external financing have lower cash holdings. Similarly, Kling et al. (2014) find a negative relationship among access to bank financing and cash holdings for enterprises in the UK. Relating to this, also Ozkan and Ozkan (2004) state that there is a negative relationship between bank debt and cash holdings for a sample of UK firms.

However, the availability of easy access to bank financing for companies cannot be taken for granted during the crisis time where the substitution effect should diminish due to firms becoming more financially constrained. Furthermore, the precautionary as well as the transaction cost motive lead to the expectation of higher cash holdings from easy access to bank financing. Empirical evidence of Duchin et al. (2010) demonstrates for a sample of publicly traded US firms that cash holdings are increased once external financing is not readily available, supporting the former prediction that the substitution effect of easy access to bank financing diminishes once debt is not readily available. Additionally, Song and Lee (2012) maintain for an East Asian sample of corporations that the negative relationship of debt and cash holdings is less pronounced in the post-crisis period of the Asian financial crisis compared to previous years.

These predictions lead to the subsequent hypotheses:

Hypothesis 3: For MNCs incorporated in developed countries, easy access to bank financing has a negative impact on cash holdings.

Hypothesis 3a: The relationship of easy access to bank financing and cash holdings is more pronounced during the period of pre-crisis compared to the years of crisis.

2.2.4. Impact of the legal system on cash holdings

The agency motive predicts that MNCs incorporated in countries with a common law legal system have lower cash holdings than corporations in countries with a code law system, to avoid agency problems as shareholders are more protected in the former system. Hence, shareholders may have the power to dismiss managers not acting in their interest from their responsibilities within the firm (Jensen, 1986). Relating to this, Pinkowitz et al. (2003) maintain, based on a large international sample, that firms with lower investor protection (i.e. corporations from countries with a code law system) hold more cash. According to them, one explanation can be that the few, large shareholders of firms in these countries aim to benefit privately from higher cash holdings. Furthermore, these findings are also confirmed by Dittmar et al. (2003) who reason for an international sample of corporations that enterprises in countries with weaker governance mechanisms hold more cash than organizations in countries with higher shareholder protection.

Contrary, evidence of von Eije (2012) for a large international sample of industrial companies emphasizes that cash holdings of firms in countries with a common law system are higher than in code law systems. In addition, also Harford et al. (2008) demonstrate that US firms with weaker governance structures, and hence lower investor protection, have lower cash holdings. Despite the previous notions, a research from Pinkowitz and Williamson (2001) concludes that US and German companies hold lower amounts of cash compared to Japanese ones. This is contradictory to the findings above due to the fact that Germany as well as Japan are both considered to be countries with a code law legal system.

Due to mixed evidence found in previous studies, based on agency theory it is expected that firms in common law legal systems hold less cash than in code law systems. Therefore, the next set of hypotheses is stated as follows:

Hypothesis 4: MNCs incorporated in countries with a common law legal system hold lower amounts of cash than the ones incorporated in a code law legal system.

Hypothesis 4a: The relationship of the common law legal system and cash holdings is less pronounced during the period of pre-crisis compared to the crisis years.

Hypothesis 4b: The relationship of the common law legal system and cash holdings is less pronounced during the post-crisis years compared to the years of crisis.

3. Methodology and data

The WRDS Compustat Global database as well as North America, to account for US firms, which are commonly used in financial research (e.g. Duchin et al., 2010; Kahle & Stulz, 2013) are employed for this study. Both files are merged, and all subsequently referred to variables can be downloaded from there. The sampling period is from 2003-2016. The beginning of 2003 is chosen to account for the time after the dot-com bubble and 2016 is set as the last year given the unavailability of financial data for the fiscal year of 2017 for some companies. A panel study is provided which allows a comparison of changes occurring within the sample throughout the whole sampling period. As panel data is used, pooled OLS regressions are tested against a Hausman test with all dummy variables, the outcome suggests the usage of firm fixed effects in each regression. However, following Song and Lee (2012) pooled OLS regressions with industry fixed effects, which are based on dummy variables of the one-digit standard industrial classification (SIC) code, are also employed. To examine the impact of cash flow, easy access to bank financing as well as the effect of the legal system on cash holdings, a modified equation based on Almeida et al. (2004) is used to also account for the differences within the various periods.

Cash holdingsi,t = α0 + β1 cash flowi,t-1 + β2 easy access to bank financingi,t-1 + β3 pre-crisis + β4 post-crisis + β5 (pre-crisis*cash flowi,t-1) + β6 (pre-crisis*easy access to bank financingi,t-1) + β7 (post-crisis*cash flowi,t-1) + β8 (post-crisis*easy access to bank financingi,t-1) + β9 country effecti + β10 (pre-crisis*country effecti) + β11 (post-crisis*country effecti) + β12 Σ(control

variablesi,t-1) + µi,t (1)

The dependent variable, cash holdingsi,t, represents the amount of cash corporation i holds at the end of year t. Cash holdings_1 (CH_1) is defined as cash and marketable securities divided by the book value of total assets (Almeida et al., 2004; Song & Lee, 2012). As a robustness test for the dependent variable another definition, namely cash and marketable securities sized by net sales, is used (Opler et al., 1999; Harford et al., 2008) which is referred to as cash holdings_2 (CH_2).

In order to tackle the issue of endogeneity which is a common problem in cash holding research due to the fact that the causality among cash holdings and, for instance, cash flow may be reversed (Almeida et al., 2004; von Eije, 2012), as well as the fact that shocks affecting the cash holding behavior may also have an impact on some of the firm specific independent variables (Ozkan & Ozkan, 2004), all following variables but the dummies are lagged by one year serving as instrumental variables (Almeida et al., 2004; von Eije, 2012; Chang et al., 2017).

All variables are described in Table 1. In order to account for currency differences since the WRDS database provides financial data in the local currency, ratios for each variable are employed and the SIZE variable is converted to Euro based on yearly average exchange rates provided by Deutsche Bundesbank (2018).

3.1. Data

To arrive at a sample, the following restrictions are applied upon the data. As this research only focuses on firms from developed countries these are defined based on the world governance index (WGI), provided by Kaufman and Kraay (2018), which takes a value ranging from minus two to positive two. The WGI consists of six different measures, namely voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law as well as control of corruption, for each country. Following Pinkowitz et

Table 1: Variable description

Variable Description

Dependent variables

Cash holdings_1 (CH_1) Cash and marketable securities over book value of total assets

Cash holdings_2 (CH_2) Cash and marketable securities over net sales Independent variables

Cash flow (CF) Operating income before depreciation over the book value of total assets

Easy access to bank financing (ATBF)

Short term debt over the book value of total assets Pre-crisis (PRECR) Dummy variable being 1 for the period of pre-crisis

(2003-2007), 0 otherwise

Post-crisis (POSTCR) Dummy variable being 1 for the post-crisis period (2012-2016), 0 otherwise

Country effect (CE) Dummy variable being 1 for countries with a common law system and 0 for the ones with a code law system Control variables

Cash flow risk (CFR) Standard deviation of operating income before

depreciation of firm i for period t-1, t-2 and t-3 divided by the book value of total assets for period t-1

Size (SIZE) Natural logarithm of book value of total assets

Net working capital (NWC) (Current assets minus cash and cash equivalents minus current liabilities) over book value of total assets Growth opportunities (GO) R&D expense divided by net sales

Capital expenditures (CAPEX) Gross capital expenditures over the book value of total assets

Listed (LISTED) Dummy variable being 1 for listed, 0 for unlisted firms

The final sample consists of 81,935 firm-year observations for the period of 2003-2016 and includes 6,775 different corporations. Table A.1 in Appendix 1 presents the industry distribution for the sample. The majority of firms is active in the Heavy as well as Light Manufacturing industry (37.48% and 18.11%, respectively). Followed by 16.15% in the Wholesale industry, 13.73% in Services, 7.19% in Mining and Construction, 6.82% in Transportation, and the least amount in Agriculture, Forestry and Fishing, namely .53%. Table A.2 in Appendix 2 presents the country distribution. Most firms are from Japan (40.61%) and Taiwan (13.36%) whereas only .01% of the MNCs are incorporated in New Zealand. Therefore, the results may be biased towards Asian countries.

4. Empirical findings 4.1. Descriptive statistics

Table 2 presents the descriptive statistics of the full panel. The average firm has CH_1 of 15.74% of total assets with a standard deviation of .1269. Comparing this to previous findings, it is higher than the 14% found by Song and Lee (2012) for Asian firms during the period of 1990 until 2006. However, CH_1 is comparable to the 15.8% discovered by Kling et al. (2014) for UK corporations in the period of 1988 until 2008. Bates et al. (2009) report rising cash holdings for US firms. In their study, average cash holdings to total assets rise from 10.5% in 1980 to 23.2% in 2006. This amount of cash holdings for US firms is higher than the amount of cash holdings investigated in this research for an international sample of MNCs. Fernandes and Gonenc (2016) report cash holdings of 16.42% scaled to total assets for a sample of MNCs during the period of 1990-2011. One explanation for the slightly higher amount in cash holdings compared to this research may be the fact that also companies from undeveloped countries are employed in their research which typically have higher cash holdings (Pinkowitz et al., 2015). CH_2 on the other hand shows a higher average, with 17.73% of net sales,

Table 2: Descriptive statistics of full panel

Variable No. of observations Mean Standard deviation Min Max CH_1 81,614 .1574 .1269 .0012 .5878 CH_2 81,680 .1773 .1754 .0008 .8572 CF 79,175 .1010 .0813 -.1769 .3609 ATBF 79,196 .0915 .1025 .0000 .4707 CFR 76,352 .0322 .0380 .0014 .2359

SIZE (in million Euro) 81,739 2,469 7,353 11.4682 52,760

NWC 79,023 .0345 .1627 -.4306 .4591

GO 74,193 .0215 .0418 .0000 .2312

CAPEX 78,215 .0435 .0447 .0005 .2471

compared to CH_1 and a standard deviation of .1754. This is lower but similar to 18% found by Harford et al. (2008) for the period of 1993 until 2004 for US firms. Based on these results, it can be concluded that cash holdings in an international setting of MNCs are on average higher than for Asian firms but similar to UK companies. However, lower than for MNCs in developed as well as undeveloped countries and lower than for US corporations in terms of CH_1. Though the amount of CH_2 in this sample is comparable to the one of US firms. The average amount of CF is 10.10% of total assets and ATBF amounts to 9.15% of total assets. This leads to the conclusion that internal cash flows are a slightly more important financing source than bank credits for MNCs in this sample, providing evidence that MNCs are financed according to the pecking order theory. CFR has a mean of .0322. The average corporation in this sample has a SIZE of about 2,469 million Euro. The amount of NWC is 3.45% of total assets and growth opportunities are 2.15% scaled to net sales. CAPEX amounts on average to 4.35% of total assets.

Table 3 presents the descriptive statistics per period as well as the difference-in-mean (t-test) and the Mann-Whitney test, of the full panel, to investigate whether the mean and median of the variables change in the different periods under investigation.

Table 3: Descriptive statistics of full panel per period Pre-crisis period (2003–2007) Crisis period (2008-2011) Post-crisis period (2012-2016)

Difference crisis & pre-crisis

Difference post-crisis & crisis

Variable Mean Median Mean Median Mean Median Mean Median Mean Median

unstable in the period of crisis but become more predictable in the post-crisis years. SIZE displays highly statistically significant positive changes, leading to the conclusion that firms become larger in both periods in terms of their total assets. When controlling for the number of firms per year, it is visible that the amount of MNCs in the sample grows continuously from 5,463 in 2003 to 6,156 in 2008. However, after this year the number of organizations decreases each year to 5,422 in 2016. Therefore, it can be reasoned that the effect of smaller firms going bankrupt during the crisis as well as post-crisis years may have an impact on the average size of firms, as well. The mean of NWC exhibits a decrease in the period of crisis compared to the pre-crisis years, thus MNCs hold lower amounts of NWC in this period but an increase in the mean and median in the post-crisis years of NWC is visible. GO depict positive changes in both periods. This leads to the conclusion that MNCs have potential to grow in all periods. The mean as well as the median of CAPEX decreases in all years, leading to the conclusion that less investments are made during the crisis period as well as the post-crisis time compared to earlier periods. Furthermore, this gives an indication that the additional cash may stem from reduced spending’s on investments that are kept as cash within the organization.

4.2. Pearson correlation matrix

Table 4 depicts the correlation matrix. CH_1 and CH_2 show a significant positive association of .7510. This is expected as CH_2 is used as a robustness test for the dependent variable. Cash flow is significantly positively related to both measures of cash holdings (.1049 and .0334 with CH_1 and CH_2, respectively). This leads to the conclusion that, as expected, organizations with higher cash inflows hold higher amounts of cash. ATBF on the other hand shows a significant negative correlation with cash holdings (-.2450 and -.2141 with CH_1 and CH_2, respectively), supporting the idea of easy access to bank financing serving as a substitute to high amounts of cash holdings. CF and ATBF have a significant negative correlation (-.2736), providing evidence that MNCs are financed according to the pecking order theory. CFR is

Table 4: Pearson correlation matrix

CH_1 CH_2 CF ATBF CFR SIZE NWC GO CAPEX CH_1 1 CH_2 .7510* 1 CF .1049* .0334* 1 ATBF -.2450* -.2141* -.2736* 1 CFR .1627* .1037* .0413* -.0165* 1 SIZE -.2096* -.0531* .1039* -.1114* -.2717* 1 NWC .0054 .0143* .0994* -.3699* .0063 -.1034* 1 GO .2593* .3607* -.0239* -.1371* .1203* .0197* .0725* 1 CAPEX -.1458* -.0499* .2533* -.0364* .1162* .0744* -.1358* -.0494* 1

positively related to both measures of cash holdings. As expected, the riskier the cash flow, the more cash is kept at hand to compensate the perceived risk (Bates et al., 2009). SIZE is negatively related to CH_1 and CH_2, thus, larger firms hold lower amounts of cash which is also foreseen (Opler et al., 1999). NWC is positively associated with cash holdings but the correlation is only significant for CH_2. This implies that the more NWC there is in a company, the more CH_2 is hold. This is unexpected as NWC can serve as a substitute to high amounts of cash holdings (Opler et al., 1999; Bates et al., 2009). One explanation for the positive association may be increased risk in the time-period under investigation and thus the substitution effect may diminish. Growth opportunities are also positively related to both measures of cash holdings. This is expected as growing firms need cash in order to finance their growth (Faulkender & Wang, 2006). CAPEX shows a significant negative correlation with CH_1 and CH_2. This is also foreseen due to the substitution effect explained above (Bates et al., 2009). The correlation among the independent variables is low. The highest correlation perceived is -.3699 among ATBF and NWC. A negative association between these variables is foreseen as ATBF is part of current liabilities which are ultimately a fraction of NWC. As current liabilities are deducted from current assets to arrive at NWC, NWC should decrease as ATBF increases.

Before running the regressions, the four Gauss-Markov assumptions are tested. The first assumption, namely the errors having zero mean is not problematic as a constant is included in each regression. Therefore, the dummy for SIC code 0 is left out in regressions with industry fixed effects. The issue of autocorrelation is assessed via a Durbin-Watson test, heteroskedasticity with a Breusch-Pagan test as well as multicollinearity with the variance inflation factor (VIF) which should be below ten (Woolridge, 2012). Multicollinearity is not an issue as the highest VIF is 6.06 for PRECR which is well below ten. The Breusch-Pagan test shows the presence of heteroscedasticity, therefore, the robust standard errors are presented in each of the following regression tables. The Newey-West standard errors also eliminate the problem of autocorrelation.

4.3. Regression results

1, 3, 5, 7, 9, 11, 13 and 15) have CH_1 as dependent variable and the ones with even numbers CH_2.

4.3.1. Cash holdings and the financial crisis

cash during these times (e.g. pursing investment opportunities once these emerge or being liquid) outweigh the costs of liquidity premiums or tax disadvantages.

The control variables imply that CFR is significantly positively related to cash holdings which is in line with findings of Bates et al. (2009). Once cash flows become more uncertain more cash is hold, to compensate the risk of foregoing investment opportunities due to the non-availability of financing. SIZE shows a significant negative impact on CH_1 and CH_2. Thus, larger firms hold less cash than smaller ones which is in accordance with evidence found by Opler et al. (1999). NWC is significantly negatively related to both cash holding measures. This is expected as NWC consists of assets which can be used as substitutes to cash holdings (Pinkowitz & Williamson, 2001). GO is significantly positively related, to cash in both models. This is also foreseen as growth needs to be financed which is easier once cash is readily available (Faulkender & Wang, 2006). CAPEX shows a significant negative relationship with both cash holding measures. On the one hand this supports the substitution effect, as assets can be used as collateral, leading to an increased accessibility of debt. Therefore, less cash needs to be hold by the corporation (Bates et al., 2009). On the other hand, it indicates that MNCs with greater investment opportunities hold less cash. Lastly, listed firms have significantly higher amounts of cash holdings than unlisted ones (.0386 and .0324 for models 1 and 2, respectively) which is contrary to the prediction above. However, also von Eije (2012) finds evidence that listed corporations hold higher amounts of cash than unlisted ones. One explanation for this behavior can be that managers of unlisted firms spend cash quickly to avoid agency problems whereas owners of listed enterprises allow managers to hold larger amounts of cash to pursue investment opportunities (Harford et al., 2008). Furthermore, unlisted firms may be more financially constrained than listed ones as the latter have better access to financing sources (Pagano et al., 1998). Thus, unlisted corporations may need to spend their cash more quickly than listed firms to remain competitive. The adjusted R² for model 1 is .2705 and .2125 for model 2.

4.3.2. Impact of cash flow and easy access to bank financing on cash holdings

access to bank financing for high amounts of cash holdings. Based on this evidence as well as the robust findings for these two relationships in models 1-8, hypotheses 2 and 3 are failed to be rejected. The dummy variables PRECR (POSTCR) support former findings, namely a statistically significant negative (positive) impact in both models. The interaction variables CF*PRECR and ATBF*POSTCR are insignificant in both models. Contrary to the prediction, no changes can be perceived in the impact of CF (ATBF) on cash holdings in the pre-crisis (post-crisis) period compared to the years of crisis. ATBF*PRECR is significant at the 10% level and the coefficients are -.0175 and -.0249 in models 3 and 4, respectively. Therefore, it can be maintained that the overall relationship of ATBF and cash holdings is more pronounced during the pre-crisis period compared to the years of crisis. More specifically, the additional effect of ATBF on cash holdings in the pre-crisis period is -.3584 (-.3409+(-.0175)) in model 3 and -.3700 (-.3451+(-.0249)) in model 4 compared to the years of crisis. Thus, hypothesis 3a is failed to be rejected since the substitution effect of ATBF on cash holdings is greater in the pre-crisis period than during the crisis. In accordance with evidence from Duchin et al. (2010) cash holdings are increased once bank financing is not readily available. The additional effect of CF*POSTCR is negative in both models, (-.0298 which is significant at the 10% level in model 3 and -.0615, being highly statistically significant in model 4). Hence, the overall relationship of CF and cash holdings is less pronounced in the post-crisis period compared to the years of crisis. More specifically the additional effect of CF on cash holdings in the post-crisis period is .1920 (.2218+(-.0298)) in model 3 and .0269 (.0884+(-.0615)) in model 4, compared to the years of crisis. Based on this evidence, hypothesis 2b is failed to be rejected as the main relationship of CF and cash holdings is less pronounced in the period of post-crisis compared to the years of crisis.

Taking models 5 and 6 into account which are used as a robustness test as these use firm fixed effects, CF is again highly statically positively related to CH_1 (.1050) but the impact of CF is insignificant in model 6. ATBF has, repeatedly, a significant negative impact on cash holdings in both models (-.1632 and -.1907 in models 5 and 6, respectively). This underlines the evidence found in models 1-4. The dummy variables PRECR (POSTCR) are significantly negatively (positively) related to cash holdings, supporting the increase in cash during the crisis and post-crisis years. Therefore, these results provide further evidence for hypotheses 1a, 1b, 2 and 3. All control variables are in line with former findings but GO. These are statistically significantly negatively related to CH_1 and the impact is insignificant in model 6. All interaction terms in models 5 and 6 are insignificant. Therefore, it can be reasoned that the results for hypotheses 2b and 3a are not robust across OLS with industry fixed effect and firm fixed effect models. Nevertheless, from the evidence in these two models, it can be maintained that the effect of CF and ATBF on cash holdings in the pre-crisis as well as the post-crisis period does not change compared to the years of crisis. Both models with firm fixed-effects have a very low adjusted R² of merely 6.78% and 4.18% (model 5 and 6, respectively). Therefore, at lot of the variance within the models remains unexplained.

4.3.3. Impact of the legal system on cash holdings

legal system. This is contrary to the evidence found by von Eije (2012) as well as by Harford et al. (2008). However, in line with the predictions of the agency motive as well as evidence provided by Pinkowitz et al. (2003) and Dittmar et al. (2003). As predicted, managers in firms in countries with a code law legal system accumulate higher amounts of funds. Therefore, hypothesis 4 is failed to be rejected. The interaction terms CE*PRECR and CE*POSTCR are used to investigate the additional effect of the legal system and the pre- and post-crisis period

Table 5: Regression results of full panel and sub-sample

Variable (1) (2) (3) (4) (5) (6) (7) (8) CF .2100*** (.0069) .0656*** (.0091) .2218*** (.0112) .0884*** (.0149) .1050*** (.0116) .0044 (.0159) .2550*** (.0091) .1259*** (.0113) ATBF -.3431*** (.0048) -.3482*** (.0067) -.3409*** (.0074) -.3451*** (.0106) -.1632*** (.0103) -.1907*** (.0147) -.4278*** (.0062) -.4533*** (.0082) PRECR -.0165*** (.0010) -.0191*** (.0014) -.0144*** (.0024) -.0168*** (.0033) -.0167*** (.0022) -.0191*** (.0032) -.0238*** (.0013) -.0246*** (.0017) POSTCR .0120*** (.0010) .0176*** (.0014) .0143*** (.0023) .0227*** (.0033) .0098*** (.0020) .0172*** (.0029) .0172*** (.0013) .0204*** (.0018) CF*PRECR -.0047 (.0160) -.0007 (.0206) -.0138 (.0150) -.0105 (.0200) ATBF*PRECR -.0175* (.0100) -.0249* (.0138) -.0043 (.0086) .0006 (.0126) CF*POSTCR -.0298* (.0154) -.0615*** (.0210) .0132 (.0142) .0024 (.0192) ATBF*POSTCR .0080 (.0154) .0116 (.0145) .0128 (.0081) .0024 (.0121) CE -.0545*** (.0024) -.0388*** (.0030) CE*PRECR .0230*** (.0030) .0229*** (.0038) CE*POSTCR -.0184*** (.0027) -.0168*** (.0037) CFR .3242*** (.0150) .2133*** (.0196) .3229*** (.0151) .2105*** (.0196) .1494*** (.0176) .0720*** (.0244) .5584*** (.0207) .3851*** (.0263) SIZE -.0178*** (.0003) -.0091*** (.0004) -.0177*** (.0003) -.0091*** (.0004) -.0198*** (.0016) -.0064*** (.0022) -.0170*** (.0003) -.0076*** (.0004) NWC -.1329*** (.0033) -.1115*** (.0041) -.1332*** (.0033) -.1120*** (.0041) -.0860*** (.0083) -.0877*** (.0107) -.1368*** (.0039) -.1113*** (.0048) GO .5530*** (.0133) 1.166*** (.0205) .5831*** (.0133) 1.166*** (.0205) -.1111*** (.0407) .0674 (.0624) .6669*** (.0168) 1.3467*** (.0255) CAPEX -.4920*** (.0099) -.2281*** (.0151) -.4926*** (.0099) -.2299*** (.0152) -.2176*** (.0127) -.2019*** (.0191) -.5413*** (.0133) -.3803*** (.0186) LISTED .0386*** (.0008) .0324*** (.0012) .0387*** (.0008) .0325*** (.0012) .2460*** (.0071) .0860*** (.0099) .0385*** (.0016) .0495*** (.0021) SIC 1 .0275*** (.0059) .0016 (.0086) .0279*** (.0059) .0022 (.0086) .0175** (.0087) -.0214 (.0133) SIC 2 .0099* (.0058) -.0314*** (.0084) .0101* (.0058) -.0309*** (.0083) -.0083 (.0086) -.0515*** (.0131) SIC 3 .0455*** (.0057) .0081 (.0084) .0456*** (.0057) .0085 (.0084) .0117 (.0086) -.0257** (.0131) SIC 4 .0132** (.0059) -.0062 (.0087) .0134** (.0059) -.0058 (.0087) .0013 (.0088) -.0332** (.0133) SIC 5 .0204*** (.0058) -.0679*** (.0084) .0207*** (.0058) -.0674*** (.0084) .0029 (.0086) -.0871*** (.0131) SIC 7 .0440*** (.0059) .0013 (.0085) .0444*** (.0059) .0020 (.0086) .0387*** (.0087) -.0079 (.0132) Constant .2265*** (.0062) .2308*** (.0089) .2248*** (.0063) .2277*** (.0091) .1554*** (.0121) .1882*** (.0162) .2503*** (.0091) .2353*** (.0137) Obs. 71,090 71,118 71,090 71,118 71,090 71,118 46,506 46,525

Model OLS OLS OLS OLS FE FE OLS OLS

Industry dummies Yes Yes Yes Yes No No Yes Yes

Adj. R2 _{.2705 .2125 .2706 .2128 .0678 .0418 .3298 .2835}

on cash holdings. For models 7 and 8 CE*PRECR (CE*POSTCR) are .0230 (-.0184) and .0229 (-.0168), respectively. All are statistically significant at the 1% level. This implies that the overall relationship of the common law legal system and cash holdings is less pronounced in the pre-crisis period compared to the years of crisis, and more pronounced during the post-crisis than in the post-crisis years. More specifically, the additional effect of the common law legal system on cash holdings during the pre-crisis compared to the crisis years is -.0315 (-.0545+.0230) for model 7 and -.0159 (-.0388+.0229) in model 8. Therefore, hypothesis 4a is failed to be rejected. The results are in line with findings from Kahle and Stulz (2013) as well as Lee and Park (2016). The additional effect of the common law legal system on the cash holding behavior in the post-crisis is -.0729 (-.0545+(-.0184)) in model 7 and -.0556 (-.0388+(-.0168)) in model 8, compared to the years of crisis. Based on this evidence, hypothesis 4b has to be rejected. The main relationship of the common law legal system and cash holdings is more pronounced in the period of post-crisis compared to the years of crisis and not, as expected less pronounced. One explanation for this behavior can be that ATBF is still restricted in the post-crisis period and firms in Japan and Germany may be more constrained in accessing financing sources even in the post-crisis period. Thus, the precautionary motive may still be of importance to these corporations after the crisis where they hold higher amounts of cash. Furthermore, agency theory may prevent mangers in firms incorporated in countries with a common law legal system to have higher cash holdings. The adjusted R² for models 7 and 8 can explain .3298 and .2835 percent of the variance, respectively.

4.3.4. Robustness test

cash holdings is therefore more pronounced in the pre-crisis period than during the crisis. More precisely, the additional effect of ATBF on cash holdings during the pre-crisis period is -.3377 (-.3231+(-.0146)) in model 9, compared to the years of crisis. Based on this evidence, hypothesis 3a is failed to be rejected. CF*POSTCR is insignificant in both models, whereas ATBF*POSTCR shows a highly statistically significant positive impact in both models. The additional effect of ATBF on cash holdings in the post-crisis years is -.2181 (-.3231+.1050) in model 9, compared to the crisis time. Therefore, MNCs in America save more cash from easy access to bank financing in the crisis as well as post-crisis years compared to earlier periods. Moreover, hypothesis 3b is failed to be rejected based on these results. All control variables are highly statistically significant and support former findings of models 1-4 but SIZE in model 10 is insignificant. The adjusted R² is .3667 and .3678 for the two models, respectively. Models 11 and 12 present the regression results for Asian firms. CF is again highly statistically significantly positively related to cash holdings and ATBF highly statistically significantly negatively. The dummies PRECR and POSTCR support the findings of models 1-8. Hence, Asian MNCs change their cash holding behavior in the long-run in favor of higher cash holdings. Therefore, hypotheses 1a, 1b, 2 and 3 are failed to be rejected. The interaction term CF*PRECR is insignificant in both models. ATBF*PRECR has a positive coefficient in both models which is highly statistically significant in model 11 and at the 5% level significant in model 12. The additional effect of ATBF on cash holdings in the pre-crisis is -.3945 (-.4475+.0530) in model 11, compared to the years of crisis. The main relationship of ATBF and cash holdings is thus less pronounced in the pre-crisis years compared to the years of crisis which is contradictory to the expectation of hypothesis 3a which needs to be rejected based on this evidence. CF*POSTCR is only significant in model 11 at the 5% level with a value of .0532. The relationship of CF and CH_1 in the post-crisis is more pronounced than in the crisis .3362 (.2830+.0532). Therefore, hypothesis 2b is rejected. The relationship of ATBF and cash holdings is also more pronounced in the post-crisis period compared to the crisis years, namely -.4709 (-.4475+(-.0234)) in model 11. Hence, also hypothesis 3b needs to be rejected. All control variables are highly statistically significant and support the results from models 1-4. Only listed firms in Asia hold significantly less cash than unlisted ones which is contrary to previous findings.

it is positive, implying European MNCs hold less cash in the period of crisis than in the pre-crisis years. POSTCR is highly statistically significantly positive, confirming an increase in cash holdings in the post-crisis period compared to the years of crisis. Based on this evidence, hypotheses 2, 3 and 1b are failed to be rejected, whereas hypothesis 1a is rejected. CF*PRECR (ATBF*PRECR) are only significant in model 13 where both are negative. Therefore, the overall relationship of CF (ATBF) and cash holdings in the pre-crisis period compared to the years of crisis is less (more) pronounced. More precisely it is.1606 (.2348+(-.0742)) for CF and -.3366 (-.2629+(-.0737)) for ATBF during the pre-crisis compared to the years of crisis. CF*POSTCR is significant at the 10% level in model 13. The additional effect of CF on cash holdings in the post-crisis period is thus .1876 (.2348+(-.0472)) compared to the years of crisis, which makes the main relationship of CF and cash holdings less pronounced during the post-crisis compared to post-crisis years. ATBF*POSTCR is insignificant in either model. Therefore, it can be reasoned that the evidence found for European MNCs is not robust. It may be necessary to control for the influence of the various countries within Europe as 21 of the 31 countries investigated in this research belong to Europe. The control variables are again in line with former results.

The regression results for Oceania, models 15 and 16 have only 415 and 416 observations, respectively. Therefore, it can be doubted whether these firms are a good representation of MNCs from Oceania, as well as the regression results in general. CF is insignificant in both models. ATBF shows statistically significant negative coefficients. Therefore, hypothesis 3 is failed to be rejected. PRECR and POSTCR as well as all interaction variables are insignificant. Based on this evidence it can be reasoned that MNCs incorporated in Oceania do not change their cash holding behavior during the pre- and post-crisis period compared to the crisis years. The control variables show similar results as in previous regressions. Only listed firms hold significantly less cash in Oceania.

5. Conclusion

This research examines the effect of the recent financial crisis on the cash holding behavior of MNCs from 31 developed countries by providing a comparison among the period of pre-crisis (2003-2007), crisis (2008-2011) and the post-crisis years (2012-2016). The final sample includes 6,775 firms from various industries.

pronounced in the pre-crisis period compared to the years of crisis and more pronounced in the post-crisis years than during the crisis. Hence, the evidence implies that there is a difference in the cash holding behavior among corporations being exposed to different legal systems and one explanation for the difference may be agency theory.

Ultimately, this research helps managers to understand the importance of saving cash during a period of economic downturn, namely the recent financial crisis. Cash flows become more uncertain and access to bank financing is more difficult to obtain. Thus, also MNCs are significantly affect by the crisis. Managing these risks in the long-term is therefore an important task for managers on a daily basis, to ensure that their business remains competitive.

Reference list

Almeida, H., Campello, M. & Weisbach, M.S. (2004). The cash flow sensitivity of cash. The Journal of Finance 59 (4), pp. 1777-1804.

Amess, K., Banerji, S. & Lampousis, A. (2015). Corporate cash holdings: causes and consequences. International Review of Financial Analysis 42, pp. 421-433.

Acharya, V., Philippon, T., Richardson, M. & Roubini, N. (2009). The financial crisis of 2007-2009: causes and remedies. Financial Markets, Institutions and Instruments 18 (2), pp. 89-137.

Acharya, V. & Richardson, M. (2009). Causes of the financial crisis, Critical Review 21, pp. 195-210.

Acharya, V. Almeida, H. & Campello, M. (2013). Aggregate risk and choice between cash and lines of credit. The Journal of Finance 68 (5), pp. 2059-2116.

Bates, T.W., Kahle, K.M. & Stulz, R.M. (2009). Why do U.S. firms hold so much more cash than they used to? The Journal of Finance 64 (5), pp. 1985-2021.

Bigelli, M. & Sánchez-Vidal, J. (2012). Cash holdings in private firms. Journal of Banking & Finance 36, pp. 26-35.

Chang, Y., Benson, K. & Faff, R. (2017). Are excess cash holdings more valuable to firms in times of crisis? Financial constraints and governance matters. Pacific-Basin Journal 45, pp. 157-173.

Denis, D.J. & Sibilkov, V. (2010). Financial constraints, investment, and the value of cash holdings. The Review of Financial Studies 23 (1), pp. 247-269.

Deutsche Bundesbank (2018). Time series databases, exchange rates. Retrieved from:

https://www.bundesbank.de/Navigation/EN/Statistics/Time_series_databases/External_sector /external_sector_list_node.html?listId=www_s331_b01012_1, accessed 24.04.2018.

Dittmar, A., Mahrt-Smith, J. & Servaes, H. (2003). International corporate governance and corporate cash holdings. The Journal of Financial and Quantitative Analysis 38 (1), pp. 111-133.

D’Mello, R., Krishnaswami, S. & Larkin, P.J. (2008). Determinants of corporate cash holdings: evidence from spin-offs. Journal of Banking & Finance 32 (7), pp. 1209-1220. Doupnik, T. & Perera, H. (2015). International accounting, fourth edition, McGraw Hill, New York, USA.

Duchin, R., Ozbas, O. & Sensoy, B.A. (2010). Costly external finance, corporate investment and the subprime mortgage credit crisis. Journal of Financial Economics 97 (3), pp. 418-435. European Commission (2018). What is an SME? Retrieved from:

http://ec.europa.eu/growth/smes/business-friendly-environment/sme-definition_de, accessed 25.04.2018.

Fernandes, N. & Gonenc, H. (2016). Multinationals and cash holdings. Journal of Corporate Finance 39, pp. 139-154.

Han, S. & Qiu, J. (2007). Corporate precautionary cash holdings. Journal of Corporate Finance 13 (1), pp. 43-57.

Harford, J., Sattar, A.M. & Maxwell, W.F. (2008). Corporate governance and firm cash holdings in the US. Journal of Financial Economics 87, pp. 535-555.

Hope, O.-K. & Thomas, W.B. (2008). Managerial empire building and firm disclosure. Journal of Accounting Research 46 (3), pp. 591-626.

Ivashina, V. & Scharfstein, D. (2010). Bank lending during the financial crisis of 2008. The Journal of Financial Economics 97, pp. 319-338.

Jang, Y. (2017). International corporate diversification and financial flexibility. The Review of Financial Studies 30, pp. 4133-4178.

Jensen, M.C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American Economic Review 76 (2), pp. 323-329.

Kahle, K.M. & Stulz, R.M. (2013). Access to capital, investment and the financial crisis. The Journal of Financial Economics 110, pp. 280-299.

Kaufman, D. & Kraay, A. (2018). Worldwide governance indicators. Retrieved from: http://info.worldbank.org/governance/WGI/#home, accessed 17.04.2018.

Keynes, J.M. (1936). The general theory of employment, interest and money. Harcourt, Brace & World, New York, USA.

Kling, G., Paul, S.Y. & Gonis, E. (2014). Cash holding, trade credit and access to short-term bank finance. International Review of Financial Analysis 32, pp. 123-131.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A. & Vishny, R. (2000). Investor protection and corporate governance. Journal of Financial Economics 58, pp. 3-27.

Lee, C. & Park, H. (2016). Financial constraints, board governance standards, and corporate cash holdings. Review of Financial Economics 28, pp. 21-34.

Myers, S.C. (1984). The capital structure puzzle. The Journal of Finance 39 (3), pp. 575-592. Myers, S.C. & Majluf, N.S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, pp. 187-221. Opler, T., Pinkowitz, L., Stulz, R. & Williamson, R. (1999). The determinants and

implications of corporate cash holdings. The Journal of Financial Economics 52, pp. 3-46. Ozkan, A. & Ozkan, N. (2004). Corporate cash holdings: an empirical investigation of UK companies. Journal of Banking & Finance 28 (9), pp. 2103-2134.

Pagano, M., Panetta, F. & Zingales, L. (1998). Why do companies go public? An empirical analysis. The Journal of Finance 53 (1), pp. 27-64.

Pinkowitz, L., Stulz, R.M. & Williamson, R. (2003). Do firms in countries with poor

protection of investor rights hold more cash? Unpublished working paper. National Bureau of Economic Research.

Pinkowitz, L., Stulz, R. M. & Williamson, R. (2015). Do US firms hold more cash than foreign firms do? The Review of Financial Studies, 29 (2), pp.309-348.

Song, K.R. & Lee, Y. (2012). Long-term effects of a financial crisis: Evidence from cash holdings of East Asian firms. Journal of Financial and Quantitative Analysis 47 (3), pp. 617-641.

The Economist (2018). The retreat of the global company. Retrieved from:

https://www.economist.com/news/briefing/21715653-biggest-business-idea-past-three-decades-deep-trouble-retreat-global, accessed 01.05.2018.

von Eije, H. (2012). What causes differences in cash holdings between listed and unlisted firms around the world? Social Science Research Network.

Appendix 1: Industry distribution

Table A.1: Industry distribution Short SIC code Industry Number of firms Percentage (%) 0 Agriculture, Forestry and Fishing 36 0.53

1 Mining and Construction 487 7.19

2 Light Manufacturing 1,227 18.11 3 Heavy Manufacturing 2,539 37.48 4 Transportation 462 6.82 5 Wholesale 1,094 16.15 7 Services 930 13.73 Total 6,775 100.00

Appendix 2: Overview of countries Table A.2 Country distribution

Country Number of firms Percentage (%) Australia 40 .59 Austria 44 .65 Belgium 51 .75 Bermuda 60 .89 Canada 79 1.17 Chile 88 1.30 Czech Republic 7 .10 Denmark 75 1.11 Finland 28 .41 France 260 3.84 Germany 243 3.59 Great Britain 690 10.18 Hong Kong 16 .24 Ireland 10 .15 Island 9 .13 Israel 109 1.61 Japan 2,751 40.61 Jersey 20 .30 Lithuania 2 .03 Luxemburg 14 .21 Malta 2 .03 Netherlands 68 1.00 New Zealand 1 .01 Norway 108 1.59 Portugal 2 .03 Singapore 11 .16 Slovenia 18 .27 Sweden 194 2.86 Switzerland 144 2.13 Taiwan 905 13.36 USA 726 10.72 Total 6,775 100.00

Appendix 3: Robustness test

Table A.3: Robustness test per continent

Variable America Asia Europe Oceania

(9) (10) (11) (12) (13) (14) (15) (16) CF .2489*** (.0253) .1366*** (.0320) .2830*** (.0155) .1533*** (.0232) .2348*** (.0189) .0780*** (.0232) .2080 (.1545) .0946 (.1524) ATBF -.3231*** (.0299) -.2691*** (.0392) -.4475*** (.0087) -.4526*** (.0132) -.2629*** (.0158) -.2721*** (.0212) -.4707*** (.1321) -.3964** (.1717) PRECR .0019 (.0068) .0094 (.0092) -.0336*** (.0030) -.0322*** (.0045) .0173*** (.0044) .0038 (.0054) -.0161 (.0386) -.0044 (.0446) POSTCR -.0088 (.0057) -.0064 (.0080) .0160*** (.0029) .0278*** (.0044) .0107*** (.0040) .0113*** (.0053) .0066 (.0345) .0200 (.0431) CF*PRECR -.0487 (.0385) -.0897* (.0485) .0148 (.0221) -.0032 (.0322) -.0742*** (.0274) -.0092 (.0316) -.1468 (.1807) -.2214 (.2125) ATBF*PRECR -.0146*** (.0464) -.1680*** (.0591) .0530*** (.0115) .0347** (.0168) -.0737*** (.0222) -.0420 (.0281) .1068 (.1626) .0390 (.2548) CF*POSTCR -.0034 (.0338) -.0295 (.0450) .0532** (.0220) .0068 (.0325) -.0472* (.0261) -.0305 (.0322) .0163 (.1905) .0429 (.2221) ATBF*POSTCR .1050*** (.0375) .1536*** (.0528) -.0234** (.0118) -.0496*** (.0180) -.0304 (.0202) .0088 (.0284) -.0344 (.1413) -.2458 (.1924) CFR .4816*** (.0323) .2488*** (.0426) .4237*** (.0232) .2590*** (.0321) .3998*** (.0253) .3901*** (.0307) .4793*** (.1503) .4027** (.1881) SIZE -.0160*** (.0009) .0003 (.0012) -.0176*** (.0003) -.0101*** (.0005) -.0086*** (.0004) .0010* (.0005) -.0153*** (.0028) -.0133*** (.0050) NWC -.1813*** (.0102) -.1742*** (.0119) -.1472*** (.0042) -.1139*** (.0057) -.1117*** (.0056) -.1012*** (.0066) -.3758*** (.0596) -.3423*** (.0739) GO .8563*** (.0272) 1.511*** (.0395) .5385*** (.0193) 1.2601*** (.0331) .4281*** (.0217) .7433*** (.0309) .8870*** (.2626) 1.5959*** (.2634) CAPEX -.3238*** (.0246) -.2535*** (.0353) -.6356*** (.0135) -.3978*** (.0215) -.2876*** (.0170) .0018 (.0260) -.4078*** (.0125) -.1421 (.1739) LISTED .0206*** (.0026) .0268*** (.0036) -.0109*** (.0013) -.0328*** (.0021) .0005 (.0023) .0087*** (.0031) -.0097*** (.0133) -.0022 (.0231) SIC 1 -.0236*** (.0077) -.0265** (.0132) -.0012 (.0147) .0353*** (.0100) .0244*** (.0093) -.0494*** (.0182) -.0485** (.0233) -.0226 (.0441) SIC 2 .0366*** (.0074) -.0120 (.0123) -.0325** (.0145) .0110 (.0097) -.0065 (.0091) -.1174*** (.0178) -.0132 (.0207) -.0650* (.0336) SIC 3 .0604*** (.0073) .0103 (.0123) -.0049 (.0145) .0419*** (.0097) .0200** (.0091) -.0962*** (.0178) -.0697** (.0270) -.1697*** (.0353) SIC 4 .0287*** (.0077) .0014 (.0130) -.0321** (.0147) .0317*** (.0102) .0066 (.0093) -.0783*** (.0182) -.0040 (.0280) .0156 (.0449) SIC 5 .0379*** (.0074) -.0614*** (.0121) -.0193 (.0146) -.0290*** (.0097) .0066 (.0092) -.1379*** (.0178) -.0560*** (.0183) -.1174*** (.0319) SIC 7 .0422*** (.0082) .0243* (.0133) .0388*** (.0147) .0629*** (.0101) .0140 (.0092) -.0891*** (.0179) -.0670** (.0301) -.1203*** (.0406) Constant .1753*** (.0105) .1426*** (.0161) .3323*** (.0150) .2762*** (.0109) .1414*** (.0100) .2007*** (.0187) .2440*** (.0394) .2777*** (.0571) Obs. 9,179 9,178 41,266 41,292 20,230 20,232 415 416

Model OLS OLS OLS OLS OLS OLS OLS OLS

Industry dummies Yes Yes Yes Yes Yes Yes Yes Yes

Adj. R2 _{.3677 .3678 .3598 .2545 .1987 .1571 .3529 .2379}